El Niño in the far-eastern Pacific: Concepts,
impacts and dynamics
Ken Takahashi, Ph. D.
Servicio Nacional de Meteorología e Hidrología del Perú
IV International Conference on El Niño-Southern Oscillation: ENSO in a warmer climate
16-18 October 2018, Guayaquil, Ecuador
1. What is El Niño? A FEP perspective
Climate in the far-eastern Pacific
28°C
26°C 28°C
26°C
26°C
26°C
PiuraChicama
Huaman & Takahashi, 2016
22°C22°C 18°C
Rainfall, SST and surface wind
Carranza, 1891
Bol. Soc. Geogr. Lima
“... Past summer, a warm southward current was
observed between Paita (5° S) and Pacasmayo
(7° S), opposite to the polar current that
constantly bathes our coast...”
“... It most likely originated in the Gulf of
Guayaquil and, therefore, had warmer water than
the ocean...”
“Thus could be explained the excessive heat of
the past summer and the extraordinary
atmospheric humidity...”
”The counter-current from the Gulf of Guayaquil
produced abnormal and excessive evaporation in
the ocean of our coast, dumping the excess of
atmospheric humidity to the land of our coast, in
the form of storms, that produced the great
floodings of April and May”
“Corriente del Niño”...
Carrillo, 1893
Bol. Soc. Geogr. Lima
Sailors from Paita (Piura) thatfrequently navigate near thecoast in small vessels, north orsouth of Paita, know thiscurrent and call it the ”Child’scurrent”, undoubtedly becauseit becomes more visible and palpable after Christmas.
El Niño chronology based on documented coastal impacts
1525-1987 (Quinn, 1992)
1982-831925-26
18911877-781828
17281720 1791
1578-79“Very strong” El Niño in red
Figure: JISAO
El Niño in the northern coast of Peru: In situ data
Piura river discharge (5°S; Feb-Apr)m3/s
Puerto Chicama SST (7°S; Feb-Apr)
1925
1983 1998
2017
°C
Very strong El Niño: SST and precipitation
Data: OISST v2, CMAP
El Niño impacts in coastal Peru (March 2017)
Trujillo (8°S)
Trujillo (8°S)
Piura (5°S)
Lima (12°S)
SST anomalies* during the strongest El Niño for Peru
* Tropical mean removed
1983
1998
Niño 3.4
Niño 1+2
1925
2017
Niño 3.4
Niño 1+2
“Basin-scale El Niño” (Warm ENSO) “Coastal El Niño”
Spatial evolution of the “El Niño” concept in the international
scientific community
1891 EN: Coastal El Niño current & rain (Carranza, 1891; Carrillo, 1892; Eguiguren, 1893)
1925 EN: Northerly wind (Schott 1931)
1957-58 EN: Bjerknes feedback -> ENSO (Bjerknes, 1966, 1969)
1982-83 EN: Precursor coastal warming (Rasmusson & Carpenter, 1982) “is not essential”
(Cane, 1983)
1997-98 EN: Niño 3.4 introduced as ENSO index (Barnston et al, 1997; Trenberth 1997)
2017 (1925) Coastal EN: extreme rain, northerly winds
(Takahashi & Martinez, 2017; Garreaud, 2018; Xie et al 2018; Takahashi et al, 2018)
2002, 2006, 2010, 2016: Types of ENSO (Larkin & Harrison, 2005; Ashok et al 2007; …)
Operacionally, countries define El
Niño according to their needs
L’Heureux, Takahashi, Watkins, et al, Bull. Amer. Meteor. Soc., 2017
• Australia (BoM): Niño 3.4, Niño 3,
SOI, ...
• Perú (ENFEN): Niño 1+2 (ICEN, East
Pacific), Niño 3.4 (ONI, central Pacific)
• EE.UU. (NOAA CPC): Niño 3.4 (ONI), ...
2. What controls El Niño effects on
rainfall in Peru?
Precipitation anomalies (%) during extreme El Niño
Basin-scale El Niño Coastal El Niño
March 2017
Wet Wet
Wet
Wet
DryDry
Data: SENAMHI
Rainfall patterns of eastern and central Pacific El
Niño SST anomalies
Eastern Pacific (E) Central Pacific (C)
Sulca et al 2016
Takahashi et al., 2011
Correlation of E
and C indices with
DJF precipitation
EOF-based SST anomaly patterns
Wet
Dry
El Niño has opposite effects on rainfall in Peru depending on
the location of the warming
Eastern Pacific Central Pacific
Lavado & Espinoza, 2014 Enhancedcoastalrainfall
Reducedrainfall in the
Andes
Takahashi et al., 2011
Correlation of E and
C indices with annual
precipitation in Peru.
Local effects Teleconnection
EOF-based SST anomaly patterns
• Strong basin-scale EN
have wet (E>0) and dry
(C>0) contributions.
• Strong coastal EN are
wet (E>0) and wet
(C<0)
Nonlinear and nonlocal relation between Piura river discharge and SST
T he very strong coastal E lN ino in 1925 11
F ig. 11 a) L inear correlation betw een the annual discharge averaged for the P iura river w ith the February-M arch SST fromH adISST 1.1 (1925-2016). Scatter plots betw een the sam e discharge and the SST averaged over b) the N ino 1+ 2 region,c) theT w region (155◦ E -175◦ W , 5◦ S-5◦ N ), as w ell as w ith d) the difference betw een the tw o (N ino 1+ 2 m inus T w ).
4 D iscussion
T he E N SO paradigm is based on the interaction be-
tw een equatorialocean dynam icsand zonalw indsthrough
SST .H istorically,how ever,the association ofnortherly
w inds in the F E P w ith E N had been noted by E guig-
uren (1894) and M urphy (1925), w hile Schott (1931)
w ent further to propose that these w inds w ere the forc-
ing ofthe coastalE N .T hishypothesissubsequently w as
countered by the finding that the coastal w inds tend
to strengthen during E N (W yrtki 1975; E nfield 1981;
R asm usson and C arpenter 1982) and W ooster (1980)
argued thatSchottfailed by “underestim ating the m ag-
nitude ofthe tim e and space scales involved”.B ut none
of this later studies explicitly analyzed the 1925 and
their failure w as in im plicitly assum ing that the sam em echanism s act in the sam e w ay in every event,despite
W yrtki’s (1975)conclusion that“E lN ino certainly does
not have only a single cause”.
N evertheless,the R C 82 and the H arrison and Larkin
(1998) E N com posites do show northerly w ind anom a-
lies during the coastalw arm ing phase, but w eak com -
pared to 1925.T he seasonality of these anom alies ap-
pears to be critical for the strong feedback betw een
SST ,the IT C Z and the northerly w ind,as the SST and
the IT C Z offP eru peak clim atologically around M arch
(Takahashi, 2005; H uam an and Takahashi, 2016). W e
argue that this feedback w as m ade m ore effective in
1925 by the dom inant cold conditions in the rest ofthe
equatorial P acific, w hich prom oted convection in the
F E P by the destabilition of the troposphere and w ith
m oist easterly advection from the A m azon. H ow ever,
strong northerly anom alies w ere also observed in early
1926,around the peak of the w arm E N SO phase (F ig
12a).T his suggests that perhaps longer-term changes,
like a low er convective threshold for convection (John-
son and X ie, 2010), could have m ade this m echanism
m ore effective in the past. In fact, the latitude of the
trade-w ind confluence in these tw o years has been the
low est in the 1920-2012 period, including the extrem e
1982-83 and 1997-98 events and there is a slight (butnot significant) northw ard trend in this latitude, per-
haps a response to the stabilization associated w ith the
long-term tropicaltropospheric w arm ing (Johnson and
X ie, 2010; Jauregui and Takahashi, 2016). C onsistent
w ith this, the low tropospheric stability estim ated as
the difference betw een the 700 hP a potential tem per-
ature from the 20C R v2 and SST in the N ino 1+ 2 re-
gion (not show n) also has a sm allalbeit not significant
positive trend. O n the other hand, the SST differencebetw een the N ino 1+ 2 and the Tw region, a stability
proxy for the F E P (see section 3.4) does not show a
clear trend (F ig 12b),although uncertainty in SST re-
constructions is an issue for trends in the zonal SSTgradient in the tropicalP acific (D eser et al,2010).O n
the other hand, m any clim ate future clim ate change
2017 20172017
Takahashi & Martínez, 2017
Linear correlation (1925-2016)P
iura
dis
char
ge(m
3/s
)
River discharge is closely linked to the zonal gradient of SST (atmos stability proxy)
3. Extreme ENSO dynamics in the
FEP region
EOF1
EOF2
ENSO diversity: Pattern and strength
Takahashi et al, 2011
DJF tropical Pacific SST anomalies in the PC1-PC2 space
Strength
Pattern
x 2016
1983
1998
1973
2010
x 1958
2005
Nonlinear response convection and winds stress E
Monthly data
Percentiles
(10,25,50,75,90%)
for E bins.
Takahashi and
Dewitte, 2016
EE
Convection
(OLR
)
Weste
rly
win
dstr
ess
SST, OLR, and
wind stress
anomaly pattern
associated with E
Ecrit
Ecrit
The nonlinearity in thewind response is likelyresponse for the growthof extreme El Niño.
Ocean advection remainslargely linear.
El Niño bimodality in the GFDL CM2.1 model
0.05
0.1
0.15 0.2 0.25
0.25
0.3 0.35
0.4
0.45
0.5
0.55
0.6
0 1 2 3 4
0.0
0.5
1.0
1.5
2.0
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
● ●
●
1923
19261930
1941
1948
1951
1953
1958
1963
1965
1969
1972
1976
1982
1987
1992
1994
1997
2002
2004 2006
2009
E l N ino peaks in observations and C M 2.1 (bivariate P D F in contours). K −m eans clusters show n w ith colors.
E
C
Moderate El NinoStrong El Nino
C
E Takahashi and Dewitte, 2016
E and C associated with peak PC1 valuesColors: K-mean clusters, Contours: Estimated PDF for CM2.1
Strong colors: observational. Soft colors: CM2.1 model
2015 (updated)
PC1 (~Niño 3)
0
0.2
0.4
0.6
0.8
1
0.5 1 1.5 2 2.5 3 3.5 4
Pro
bab
ility
de
nsity
Peak T (degC)
Obs
1972
1982
1997
2015
El Niño bimodality suggested by obs (Niño 3)
Takahashi, Karamperidou and
Dewitte, 2018
Nonlinear wind
stress response to
surface warming
(Bjerknes
feedback)
Takahashi, Karamperidouand Dewitte, 2018
1997-98
2015-16
Recharge-discharge model with convective nonlinearity
Takahashi, Karamperidouand Dewitte, 2018
, a < 0
Once T (Niño 3) exceeds Tc=1.5K (~ 27.5°C), the damping on T is set to
zero (globally stable).
This results greater growth of strong El Niño events, leading to a
bimodal distribution.
Requires sustained external forcingfor the onset of such events.
Evolution of ensembles in the nonlinear RD model with El
Niño-favorable initial conditions
Tc
T
T
Solutions from the Fokker-Planck equation (same results from numerical ensembles)
Takahashi, Karamperidouand Dewitte, 2018
h
The external(stochastic) forcingproduces ensemblespread, allowing partof it to exceed Tc, leading to a separatemode evolving as strong EN, while therest continues to evolve as moderateEN.
In this model, externalforcing is essential forstrong El Niño events.
4. Extreme coastal El Niño dynamics
Daily SST at Puerto Chicama (7°S) at the coast of Peru
Onset• Several months for basin-scale EN (1982-83, 1997-98)• Days-weeks for coastal EN (1925,2017)
Shallow ocean warming during the coastal El NiñoP
rofu
nd
idad
(m)
ENFEN, 2017 (IMARPE)
Pro
fun
did
ad(m
)
Garcés-Vargas et al, 2005More than 100 mLess than 30 m
Subsurface temperature (°C) anomaly at Paita (5°S)
Subsurface temperature anomaly (°C) near the coast at 2°S
1997-1998 basin-scale
El Niño
2017 Coastal El Niño
Coastal El Niño is associated with northerly wind anomalies
and strengthening of southern branch of ITCZ
March climatology March 1925 March 1891 March 2017
Fast meridional dynamics (Schott 1931), probably involving the wind-evaporation-SST (WES; Xie and Philander 1994) feedback. Minor role of equatorial dynamics (Takahashi &
Martinez 2017; ENFEN 2017; Takahashi et al 2018)
1. The definition of “El Niño” events originally referred to an
abnormally warm ocean and heavy rainfall in the northern coast
of Peru. As such, 2017 was arguably the third strongest El Niño
on record.
2. Two very distinct types of such strong El Niño take place in the
FEP: extreme basin-scale (ENSO) events (e.g. 1982-83) and
coastal El Niño events (e.g. 1925, 2017).
3. El Niño rainfall impacts are of two kinds: wet coastal anomalies
due to eastern Pacific warming, dry conditions in the Andes due
to central Pacific warming. Individual El Niño events combine
these two elements differently.
4. Strong basin-scale El Niño grow further as eastern Pacific
warming exceeds the threshold for deep convection, which
amplifies the Bjerknes feedback.
5. Strong coastal El Niño have fast meridional dynamics associated
with the strengthening of the southern branch of the ITCZ and
shallow warming.
Conclusions
Servicio Nacional de Meteorología e Hidrología del Perú –SENAMHI
Jirón Cahuide 785 – Jesús María, Lima -Perú
Teléfono: (01) 6141414
Consultas y sugerencias: [email protected]
